Speaker 1 00:00:03 The trick of working in any system is to recognize which particular parts of what you find are idiosyncratic and which parts are hen sad or ways into general principles. And I think the most effective scientists are the ones who really have learned to distinguish between preparation and single secrecies and big picture ideas. The problem that most preoccupies me now is trying to understand how all these part of patients and an animal sees or goes through how they leave traces. That can be completely encrypted until you stress. It does. It means you're getting perfectly normal looking rhythms. Um, but there are things that have changed that you don't reveal until you give another strong probation and you realize it's different.
Speaker 3 00:01:07 This is brain inspired.
Speaker 4 00:01:21 Eve martyr runs the martyr lab at Brandeis university, where she studies the modulation of neural networks. She's my guest today. Hi everyone. I'm Paul Eve's name and work almost always comes up when discussing the challenge of understanding how large networks of neurons or units in the case of deep learning produce the cognitive functions and behaviors. We're interested in explaining in terms of those networks. The reason why her name always comes up is because of her work on a rather small network of neurons. It happens to be in the stomach of crabs and lobsters, the stomata gastric nervous system. These are networks of about 30 neurons with the types of neurons and their connections well mapped out and the neurophysiological properties of the neurons well understood. And this STG network produces rhythmic or oscillatory outputs important for the life of the crab or lobster. And the network manages to function under a wide range of sets of parameters under a wide range of environmental conditions.
Speaker 4 00:02:27 One thing this means is that nervous systems are robust and resilient, which of course is a good thing, but it also means it's nearly impossible to look at a network and infer how it's producing its output. And if that's the case for a network of 30 neurons, what hope do we have with networks of millions of neurons? So her work, for example, raises concerns about the time and energy and money being spent to map out connectomes the structures of neural networks in the hopes that the structure will tell us all about its function. As Eve says, the structure is necessary to understand function, but absolutely insufficient. So we discuss that. We discuss factors that contribute to network resilience like homeostasis, plus how, even though the networks are resilient, they start to fail in different ways when they're pushed into different modes and how this could be bad under conditions of climate change or trauma and other challenges that cause underlying changes in the network that we can appreciate until it's too late, you can learn more in the show notes at brain inspired.co/podcast/ 130 and on the website, you can also choose to support the podcast through Patrion, uh, and join the discord community where we discuss topics from the episodes.
Speaker 4 00:03:43 And we're also having more regular live zoom discussions about neuro AI, uh, type topics. So check that
[email protected] I'd been wanting to have Eve on for a long time. So I was grateful to finally be in her presence. Enjoy How much of a problem is it for? Well, first of all, do you consider yourself a famous neuroscientist?
Speaker 1 00:04:06 Um, I'm probably a moderately famous neuroscientist
Speaker 4 00:04:13 Or
Speaker 1 00:04:13 People who are more famous than I am. And there are people who are not as famous as I am within certain circles. I'm probably pretty famous, but in terms of the lay world, I'm probably not so much
Speaker 4 00:04:25 Well. Right. Would you consider it within the neuroscience world? Would you consider yourself or prefer to consider yourself famous or legendary if you had to choose?
Speaker 1 00:04:37 I wouldn't want to choose has a lot of meaning.
Speaker 4 00:04:43 Yeah. Well, does that kind of status though? I mean, that's chosen for you I suppose, but, um, you know, you're just incredibly busy and with that kind of status comes lots of responsibilities. Did you ever want all these responsibilities of having to do silly podcasts like you're doing right now, et cetera?
Speaker 1 00:05:03 Um, hopefully this will not be a silly podcast, hopefully an educational podcast. I do these things now in the hopes of maintaining some amount of excitement and imagination and optimism and the younger generation. Um, that's precisely why I do them. I hear from young students or from students fairly often saying I heard this and it made me want to do neuroscience. So that's why I did
Speaker 4 00:05:41 Well. That's good. You're going out to the right people today then hopefully. So, uh, you've done tons and tons of work on the crustacean stomata, gastric nervous system. Um, but before we, and we're going to talk a little bit about that work and the implications. And so on before we do, though, I'm curious, um, how you would characterize how your interests have changed over your career. Did you begin thinking that you were going to be working on a relatively quote, unquote simple, um, part of the nervous system like this, or, uh, did you always have, um, or did you start off with a broader picture in mind and then like many, uh, scientists, the questions become narrower and narrower, although your S you know, the implications of your work are gigantic.
Speaker 1 00:06:28 I actually think I've been fairly steady in terms of the kinds of problems that interests me. You as an undergraduate, I was quite fascinated by what was then the popular topic of, um, the innovation supersensitivity, which looks at the role of, um, activity and neurotransmitter and receptor number and localization. So I always was attracted to things that the interface between activity, um, cellular molecular regulation and synapses and neurons, there was that muscles. Um, I, I have a fairly well-developed sense that we can come back to this later, but I think it's, uh, a very important theme, which is I have to work on something which has enough or a little enough ambiguity so that I can understand what the problems are. And I think in neuroscience people, I like to sometimes think about people as fractionating themselves on ingredients of how much ambiguity they can tolerate. And there are some people who can only work on single proteins and other people can only work on consciousness, and we all sort of checks the level that's most comfortable. And, um, I run into problems when I start thinking about large numbers of their own. I don't have to worry about them
Speaker 4 00:07:59 Well. Okay. So you're more comfortable with, with a smaller degree of ambiguity, but some of the implications that your work has, um, revealed seems to put you in a higher realm of ambiguity given, and, you know, we'll talk about what those are, but do you feel like, uh, I mean, it's still like a 30 or so neuron system structurally that you're, that you're working on,
Speaker 1 00:08:22 But no, there's a difference between the ambiguity of what you're doing and understanding the general principles that you find. So my goal has always been to use a small number of neurons, um, that can be identified recorded from those dynamics can be described carefully, and then use those to try and look for general principles. Um, because I think often when you have large numbers of neurons, it can be very difficult to see the general principles because they're hidden in all sorts of other complications. So the trick is to, is to work at a level of ambiguity that still has enough mystery in it, so you can learn new things. Um, but that is well enough to find, see who find those general buttons. And I just like to say, going forward, the trick of working in any system is to recognize which particular parts of what you find are idiosyncratic and which parts are sort of sad or ways into general principles. And I think the most effective scientists are the ones who really have learned to distinguish between preparation, idiosyncrasies, and big picture ideas.
Speaker 4 00:09:48 Very good. Well, let's, let's talk a little bit about what you do. So, um, you have fishermen who will not tell you where they fish, uh, deliver crustaceans to your lab. Why is this just a fishermen thing? Because they don't want to get it's a competitive industry and they don't want to give away the location
Speaker 1 00:10:05 And the fishermen never tell anybody where we'll catch, right? So the fishermen go out into the ocean and they're very brave. And they've been doing this very difficult, hard job for a long time Fishing for any
Speaker 4 00:10:28 Don't. They use cages though. And then, then they'd just go bring up the cages or,
Speaker 1 00:10:33 And go out in small boats in the, and the dark and the waves and the winds and the storms. And they, they are incredibly brave to fight the elements. And often they have to work very hard for not much return. So yes, I have total admiration haven't you actually ever seen the movies of the storms, but battering the fishing boats and people hanging on for dear life are being blown overboard.
Speaker 4 00:11:05 It's just Hollywood, right?
Speaker 1 00:11:09 And actually in the north Atlantic, on the Seacoast, in Glossier and up to the coast of Maine and down through Bedford and all of these seagoing villages where fishing was a time-honored profession, um, there are many people who were lost. I mean, it's not uncommon that boats are lost. And the coast guard actually rescues a fair number of people from half sized vessels. So anyway, on the happy note, the fishermen go out and then they bring their catch in. And eventually, um, the catch goes to some sort of distributor. We don't buy them as active for the fishermen, but we buy them from just super distributors in our case, one of them right on the waterfront in Boston, right outside of Chinatown. And we get them from there. And in, in usual days, we call them up and they pack them in a box and call a taxi. And the taxi brings them out to the lab because we're worried about 10, 12 miles due west Boston in bad days, we go and get them from the distributor.
Speaker 4 00:12:22 Oh, I see. So what, why is it important to, to, well, have you considered, uh, having a facility where you raise them and breed them? I don't even, I don't know anything about breeding and raising crustaceans, but It's tough, huh?
Speaker 1 00:12:39 Yeah. Clearly the, um, the lobsters legal, the minimum size lobsters are five to seven years old. And the crabs that we use, which are also adults are also a multi-year old lobsters for a while. We used to raise from embryos to small juveniles in the lab. They're incredibly difficult to raise. Um, crabs are pelagic, which means they swim freely in the ocean. So I don't know of anybody who's actually raising crabs in a laboratory environment.
Speaker 4 00:13:15 So I naively thought that that your primary reason for, for doing it that way was so that they would have an ecologically valid life essentially, and have been exposed to natural environments. Okay, good. I will.
Speaker 1 00:13:30 Um, but that's, that's one obvious advantage of working on wild caught animals. They've had a full rich and probably difficult for some of them life out there in the wild. Um, but additionally, it's, it's also, they're slow growing animals. And so it's not a fun thing to imagine raising enough animals that take five to seven years to, to reach full size before you study them.
Speaker 4 00:14:03 I don't even know what would be more expensive. It seems like a lab, keeping them in a lab would be more expensive perhaps than just purchasing them. Yeah,
Speaker 1 00:14:12 No doubt. It's it's not even in the same ballpark.
Speaker 4 00:14:15 Okay. Okay. Yeah. So, all right. So, um, so you get these crustaceans lobsters and most recently crabs mostly, right? Or do you still,
Speaker 1 00:14:23 We work on both, but yeah, but a lot of the stuff we do now is on crabs
Speaker 4 00:14:28 And then you prepare their stomata gastric nervous system, which means you dissect it out. Right. Uh, and, and the reason you like that preparation is because it has around 30 neurons. And like you were saying, you like low ambiguity and it's a very well-defined system. And this is sort of a, um, an oscillatory rhythmic system involved in digestion, right?
Speaker 1 00:14:51 It's an example of what's called a central pattern generator. That's a group of neurons that produces rhythmic motor patterns. And in the case of the semantic aspect, nervous system, the ganglion itself has about 26 neurons and crabs without 30 neurons and lobsters. And each neuron can be individually identified on the basis of its projection to the muscles of the stomach. And it's the ease of identification and the ease of being able to record from all of them simultaneously that makes it such an ideal preparation.
Speaker 4 00:15:26 Do you worry that, um, I mean, there's a lot of, uh, concern now or hype around having ecological validity in scientific preparations. Right. Um, and so when you dissect that out, do we consider it still ecologically valid? Uh, even though it's not part of the rest of the functioning organism?
Speaker 1 00:15:47 Well, they've been since this preparation was first studied in the 1970s, late sixties, early 1970s, some of the earliest work involved putting electrodes in the intact animal in the muscles, and then later the nerves. And it turns out that the motor patterns that are produced and they intact animal resemble very, very closely what we see in the sector's nervous system. So the other reason why this preparation was so attractive, um, 50 or 60 years ago was precisely because the deceptive victim motor patterns as they're called, are so close to the actual motor patterns in the alive animal, that people found comfort that they could study the mechanisms that were likely relevant to what was happening in the animal, just because the patents resembled. So resemble the two sets of patterns resemble each other so closely.
Speaker 4 00:16:49 And do you still agree with that?
Speaker 1 00:16:51 Absolutely. Absolutely. We've gone back every now and then to do work in vivo and nothing has given us cause, um, most recently in 2014, we published a paper that involved, um, in vivo recordings and VHR aquariums and comparing the two and they were very, very
Speaker 4 00:17:13 Okay, good. Yeah. So let's talk about, uh, the conundrum that you have thrown neuroscience into, uh, with these long series of experiments that you've performed. So, uh, in principle, um, neuroscience has long thought, right? That as long as if we have a circuit wiring diagram, we can say something about the function of the circuit and I'm, I'm going as big picture as I can here. And we can narrow it down as we go, I suppose. But a large part of what your results suggest is that you can't look at a wiring diagram and say almost anything about its function. Would that be an accurate summary?
Speaker 1 00:17:55 Um, not quite. I would turn it around a little bit and say by 1980, we had a wiring diagram for this mini gastric nervous system and people, and they were wiring diagrams for a number of other small nervous systems, central pattern generators. And it was pretty clear that they were all different, even though some of the motor patents they produce pretty similar. Some of the initial early workers were very upset because they thought that there'd be a sort of generic central pattern generator, con connectivity that would work for all animals. And they were sort of disappointed with that turned out not to be true. I never expected it to be true, so I didn't understand why they were disappointed, but I think what we've been going through in the 40 or something years that we've been post ConnectTo, um, if you will, we have learned that the connectome or the wiring diagram is we used to call it is absolutely necessary and completely insufficient. So you can't understand how a signal works without a wiring diagram because the wiring diagram, even though it doesn't necessarily tell you exactly how it's going to work, it constrains the kinds of things that can arise. So, um, you know, it says here, the wind diary gives you a set of possibilities, and then you have to understand how those possibilities get turned into actually, um, because not anything is possible, not everything's possible,
Speaker 4 00:19:44 But in a system of 30 well-defined neurons connected with, with well-defined levels of neurotransmitters ion channels within some range, although it's a fairly wide range, I understand, uh, that, that set of possibilities is vast, right?
Speaker 1 00:20:01 Yes and no. You know, it's, it's always the case that, that, um, there are lots of different motor patterns that can be produced, but they're all explicable. And I think that's, that's the key feature. They really don't transgress. I mean, when, when you start seeing motor patterns that are enough different, you know, that something's broken,
Speaker 4 00:20:32 That's often what you spend your time doing is trying to crash the system, right. To push it to its limits.
Speaker 1 00:20:37 Right. Right. And the system is extremely stable. That, I mean, what it's actually fairly hard to crash
Speaker 4 00:20:46 Is that, or is that, uh, optimistic and hopeful and amazing
Speaker 1 00:20:51 And helpful? I mean, it's, it tells you that even though each animal has a different set of conductance densities and a different amount of transmitter, and then wherever that they found a solution, that's good enough for the animals to live out there in the world. And that means that every human, every healthy human has a different brain, but those brains generally allow most of us to breathe and to walk and to run. And hopefully the hear and see, and think even though, and you can do that with many different solutions. So I think it's, it's very gratifying to know how the difference is, do not preclude very similar behavior.
Speaker 4 00:21:35 So one of my listeners wanted me to ask you what it is that we actually can take from structure. Uh, if it's not simply, we can't look the diagram and think that's the function that would come out of that structure. And what you're saying is we, we can constrain the space of possibilities.
Speaker 1 00:21:53 Absolutely. You can constrain the space of possibilities. We don't exactly know how that's constrained, but you constrain the space of possibilities and you make some outputs far more probable than others. So if you have reciprocal inhibition, two cells that each inhibit each other well, probabilistic likelihood is that they're going to be firing an alternation or out of phase. Now they can sometimes fire at the same time, but often they won't.
Speaker 4 00:22:27 W would it be fair to say that there then are motifs that it's a space of possible motifs that we would, is that a way that we can constrain what's possible?
Speaker 1 00:22:36 So many years ago, Peter getting who was the then leader in the field of small nervous systems in the mid eighties, he was working on a molluscum China Tonia, and he was looking for what he called building blocks, circuit building blocks. So he was looking for either principals at the single cell level or the small circuit level that would have, that would sort of create the library of mechanisms then could be combined to, to build circuits with many different behaviors. But you would expect that, you know, the transit outward current when usually do X or Y or bursting, there could give you certain kinds of behaviors, et cetera. And so I think he wrote a very beautiful paper in a way when at 95 and then one in 1989, where he articulated this concept of building blocks, which was his way of articulating the, kind of almost the alphabet of how circuits could work and nothing that's been done since this changed the, what said in 1989 or the way his way of thinking. I mean, people are still using similar kinds of reasoning as they look in the small and large circuits.
Speaker 4 00:24:06 So that's going from structure to function. Is there anything that we can say going the opposite way? If you see a function, can you say anything about the structure? So structure is necessary, but not sufficient to understand the function. What about the, uh, opposite way? I mean, I suppose if you see a rhythmic pattern, there must be a central pattern generator like structure, right?
Speaker 1 00:24:30 Yeah. But that could be all different kinds. You don't necessarily know how that's built. Um, I think it's probably that inverse problem. That's probably very hard. I think I wouldn't want to try and go in that direction. People, I think that there's a route to a lot of mistakes going in that direction.
Speaker 4 00:24:55 Okay. That's almost a more theoretical direction, I suppose.
Speaker 1 00:24:59 I mean, theorists would do that. They would say what would be the likely circuit? I mean, you could take that theoretical study. You can say, here's the behavior. Now find a circuit that will give you that. And then you could use some sort of genetic algorithm to try and find the best fit to using that. And different people might find very different structures that there couldn't be very different circuits. And then, and then you're forced to say, well, which is the actual one. If you care about that, and this, this goes back to the whole issue of why people do theory and what they're trying to achieve. If you're trying to build a brain for a robot, it doesn't matter how a nervous system actually does it. If you're trying to understand how a nervous system does it, then it matters a lot how it does it.
Speaker 4 00:25:46 Well, it depends on the kind of brain, the kind of robot you want to build. Right. If you want to build AI and AGI robot, it might matter. Or maybe it might not. We don't, we don't know. Right.
Speaker 1 00:25:56 Yeah. I don't know. I mean, I think if you only want to build a robot to do a task, it's not clear to me that you have to be defined by neuro biologically plausible roles.
Speaker 4 00:26:08 How has theory, uh, affected you in some sense, what you do is very bottom up, right? Even by, you know, like running millions of simulations to see what, um, sets of parameters might give rise to similar behaviors, that's kind of a bottom up approach, but has theory, um, how much has theory, uh, inspired your work? Or how much do you, do you dabble in the realm of, of theoretical principles? Because I asked, because these days in neuroscience, everyone's saying we need better theory. Right? That's what we're missing is great theories. And I'm talking, that's generally about, you know, the brains writ large right. Brain function, grand theory of the brain, et cetera. Um, but I'm wondering if that applies to you
Speaker 1 00:26:56 Into that. I think, I think what they can give you are very clear answers, how you might build the nervous system to do a given task or set of tasks or the switch between a bunch of tasks. Um, I, I personally don't think we should be looking for the grand theory of the brain. I think we could be looking for a really good model for that explains how the basal ganglia or I think we could be looking for, um, a pretty good model for how the cerebellum works. I think you'd be looking for, um, a really good model for a lot of things, but I just don't now I'm sure that people would disagree with me. I know there are people that disagree with me, but I just don't know what a theory of though, grand theory of the brain, what, what that would tell you or what it would be because the brain is doing so many different things to so many different parts, they're all connected to each other, but you know, the rules that you need to get the respiratory centers to work, right.
Speaker 1 00:28:03 And to be stable and robust are different from the kinds of rules that you would need to get quartets to store a memory. I mean, you might have some of the same molecular components and you probably have learning in the respiratory center as well, but it's just not clear that there'd be some global theory that would give you both respiration, you know, memory of your grandmother from 50 years ago. That said there are some basic pieces of biology that go across, but those tend to be the, the more, the more basic parts of the cell biology of neurons that are totally general. So for example, if you want to understand what potassium channels do, um, they can do a lot of things, but chances are, they can be used to do a lot of things in any neuron that you want to place them. I mean, they, they, they function that way.
Speaker 1 00:29:07 So, um, but that's not what the people who are talking about the brain you're talking about, they're not telling you about what is a potassium channel doing to electrical excitability. But the thing that we've done that I think is closest to a grand principle, um, it goes back to the, the early nineties when, when we were trying to come up with models to explain the dynamics of single neurons and with when Ella was on, um, we came up with what became the first homeostatic model to explain intrinsic excitability, intrinsic excitability. And to me, the problem was very, very clear as a biologist. We were, we already had some measurements of conducted stance studies, and it was really hard to build a single neuron model that would actually capture the dynamics of the single neuron that the data came from. And we realized that to get the model to work, right?
Speaker 1 00:30:13 You have to be tuning, not just one kind of ion channel, not two kinds of iron channels, but six or eight kinds of iron towels at the same time. And so it became very clear that no one was thinking about the mechanisms that you would need to use to coordinately tune the numbers densities of all the channels in a cell. And so that original homeostatic model, which said the neurons had a target excitability level, and then they couldn't regulate the number of ion channels and the membrane accordingly as they drifted away from their target activity. Pattern that to me is an example of what I think is a big general principle that every cell in the brain has to have to solve this problem of how to control its intrinsic excitability. It doesn't mean that all cells have to use exactly the same roles. Um, but it does mean that if you want to sell to have a certain characteristic buying pattern, that there have to be ways that it can regulate its channel densities to do that.
Speaker 1 00:31:22 Um, so that's an example. We were trying to build a computational model of an LP neuron in the, in the crab. And we ended up with what I think is one of the big, big picture insights, um, in science or a science. And if you say, and I believe that any neuron anywhere has to, you know, has to solve this problem. And as I said, it doesn't necessarily have to solve the problem in exactly the same way that it has to solve this problem. It's just like every animal has other big picture issues it has to solve
Speaker 4 00:32:02 Is this where homeostasis, uh, plays a large part.
Speaker 1 00:32:05 That is the homeostatic regulation of intrinsic excitability. We didn't call it that at the time we called it activity dependent regulation. Um, but that it was the first real homeostatic model of obviously want to talk about homeostasis of the level of the animal physiology, but we just took those the negative, the concept of a negative feedback and a control loop to the single moon.
Speaker 4 00:32:36 Some people think speaking of a large theories of the, some people think that neuroscience has it all wrong. Um, and that essentially what the brain writ large is, is a, is one big hierarchical series of control loops, and that we should be applying control theory to study everything. So this would accord with that, at least on the single neuron level essentially.
Speaker 1 00:32:59 Well, but I mean, I think that's, I don't think there's any intrinsic conflict, certainly the brain like the body has multiple loops within loops. And, you know, the, the fascinating thing about biological systems is that they manage, but loops within loops within loops within loops, without it crashing all the time, which is not so trivial. If you start, you know, just doing it as an engineer,
Speaker 4 00:33:30 But, but that is what gives rise to the resilience, even in something like the STG. Right?
Speaker 1 00:33:36 Right. That's the resilience comes exactly from understanding those sorts of control loops at the lowest level, and then seeing how, how you don't do anything that comes into conflict with those loops. But, you know, I think I'm not saying that control theory is the be all end all. However, it is certainly true that biological systems manage to use these sorts of negative feedback control loops very effectively over a large scales in time in space. And I think biological systems do it much better than an engineer's do yet,
Speaker 4 00:34:30 Yet. Oh, D but you're leaving the future open for the engineers, huh?
Speaker 1 00:34:35 Yeah. I mean, they, they they're smart. And the thing is, in the beginning, they were, they were crippled by the fact that the computers they had to work with had limitations. And, you know, right now they're much less limited by the computational size and scale of what the simulations that can move. So the future lies ahead.
Speaker 4 00:34:58 I'm thinking about technology since you mentioned it, um, let's talk about, for whatever example you want, let's say the basal ganglia right. Of a higher primate is, would there be, let's say so, like the dream is always, you can record from all neurons, right? We can't, we don't have this technology yet, but let's say in 20 years or something, somehow we can, would your approach, would you recommend your approach? Um, scaled up to something that massive and, uh, interconnected.
Speaker 1 00:35:30 So here's where I have a failure of imagination Because if someone were to give me recordings from a million cells and the basil, I wouldn't know what to look for.
Speaker 1 00:35:47 So now there's some very smart people who might run all sorts of different data analysis algorithms and this, that, and the other thing, but what you're looking for, it'd be very hard to see who the raw with your eyes. And I may be enough of an old fashioned old lady that here's where I say I have problems with the ambiguity that comes with large numbers. Um, if I look at the raw data from any of our preparations, I can always see it in the raw data. And then we do do it many, many times that you get complicated, their analysis and whatever, but at the end of the day, I have to recover something I can see in the raw data and in recordings from those data. I don't think I would have the ability to see anything in the raw data. So I would never know if you got it right.
Speaker 4 00:36:45 Well, what's the raw data in your case, are you talking about your favorite graph of 2020 with the ionic? Uh, uh, conductances
Speaker 1 00:36:54 That would be one thing or the raw data would just be the spike trends. I consider Rhonda and the actual physiological output recordings from them, from the system that shows you what it's doing and when it's doing it. And so I can look at the recordings from the 30 neurons, Madagascar ganglion. I can see if I have eight electrodes or so I can see them all work at the same time. I can see the relationships in my room. I might have to use some data analysis to look at in detail at the correlations by ever, but I can see,
Speaker 4 00:37:33 Oh
Speaker 1 00:37:35 Yeah, I can see it all. Um, and there's no way I'm going to be able to see those relationships in those million neurons recorded. Even if they had tags on them that told me what kinds of neurons they were. So, you know, minimally, you need to have them tagged. So you knew who was who, and that's, that's difficult, but also even if you knew who was who and, you know, and you know, other people, you know, people who are 30 years old now, they're going to say, oh, we'll be able to do this. And maybe they will, but I don't, I don't see it for me. And that's where I said, you know, I get stuck. I get stuck when I try and go to large ensembles and many neurons, because I'm never gonna, I don't know how I would know that I've gotten it. Right.
Speaker 4 00:38:28 How has your work, uh, on the STG, has it changed your viewpoint or changed your mind about any higher cognitive functions or has it given you any insights into any higher cognitive functions, or is that something that you prefer to, that you're uncomfortable with the level of ambiguity there as well?
Speaker 1 00:38:52 No, I think there's some things that I don't know, the work I work on the STG, but I know several things that is that, I mean, we know from psychology and we know from neural network theory, that all memory is reconstructive. And that I think is a really important principle that came from actually theory first. Um, we know from all of the cell biology of neuron, now, our work in other people's work that every iron channel receptor turns over with in minutes and days and weeks. So we have a vision of the brain as constantly rebuilding itself. And the minute you think about the brain is constantly rebuilding itself. And you think about memory and cognitive processes, an emergent process of a brain that's constantly in a state of flux. Then you sort of say, there's this, um, really, really interesting set of problems as you think about a brain that's constantly having to reveal to sell, but also having to constantly maintain stable function and long-term memory.
Speaker 1 00:40:09 And so it's a fast, that's the conundrum to my, to my mind, that's the conundrum. And I think we see it far more clearly to me. I see that conundrum much more clearly now than 50 years ago, but because we know so much more about cellular mechanisms and we know so much more about how memory works, but I, I think that that really, to me, is the big, big mystery, how you have a brain, which is rebuilding itself. Totally. You know, there's probably not a single potassium channel in my brain today that was there 30 years ago. Um, and, and yeah, you know, I can remember wearing north, south, east and west on when my mother handing, walking up and down Broadway on the upper west side, right. She taught me that uptown was north and downtown was south and Hudson was west and central park was east. And you know, that, that was with me when I was three years old. And it's with me now,
Speaker 4 00:41:16 Is that what you picture when you treat, when you orient yourself, you picture the avenue and the,
Speaker 1 00:41:22 She taught me the proof of maps. I mean, she taught me orientation maps with central park and, and, uh,
Speaker 4 00:41:33 But at that, but that potassium, uh, channel has gone,
Speaker 1 00:41:37 That was long gone. And that map is still there. And so to my mind, and there was a specific incident that my mother taught me that, right? So I have a memory of her teaching me this and that memory. Now, obviously that memory is somehow other stored in cells, completely turned over all their cellular constituents. Although many of those neurons are probably still there. So that to me is the one of the big imponderables, um, why that was such an important memory that I think she was walking with me in the middle of the summer, I would say,
Speaker 4 00:42:17 And still very vivid, But you're reconstructing it right now.
Speaker 1 00:42:21 I am, but I think I'm probably getting it approximately, correct. Unlike many memories, which are not probably right.
Speaker 4 00:42:33 So I mean this, in some sense that it must give you your own research then must give you hope. Um, so, you know, thinking about the resilience of the STG system and, uh, its adaptability in different environments, right, where you can get the same rhythm or very similar rhythms, or, you know, even if they're different, there's still functionally rhythms, uh, out of very different, uh, setups, very different sets of parameters. So th that must, in some sense, give you hope that we will be able to have a handle on this sort of thing using that kind of same principle.
Speaker 1 00:43:08 Um, that's the assumption. Yeah. So it has to be that way, right? Because you can't have, By definition, you have how many billion neurons in your brain, and most of them are the ones that were there that you were born with. And none of the molecular components are really the same. And you know, it's not an accident that that's your brain and your heart that have these long loop cells that, that have the same basic problem of maintaining constantly, you know, in your heart, again, doesn't turn over cells, but it has to turn over time and channels. And it has to maintain constant function.
Speaker 4 00:43:57 It's terrifying to me every time I think about how my heart can never stop or else I will, you know, but it really goes for a long time. Well, you know, it can't okay. Yeah. Glenn,
Speaker 1 00:44:09 Your heart has just to remind you, your heart has so many ways of maintaining its function. They do a very good job. Speaking about degenerate mechanisms. There are many degenerate mechanisms that keep your heart working
Speaker 4 00:44:28 Well. That's, that's good to know. I appreciate you soothing me.
Speaker 1 00:44:31 I'm not worried.
Speaker 4 00:44:33 I'm a little worried about it, but that's okay. I'm aging. You know, one of the things that you have found and among the many obviously is in trying to essentially crash the, uh, STG system you've taken, um, crustaceans from warm winters, colder winters, and the variability in the environment, uh, leads to a different range of adaptable solutions, right? In sort of a, Well, I know that you're concerned about the environment, but go, go ahead. What?
Speaker 1 00:45:07 Well, actually the first time in 2012, when Sarah had that, got our first set of recordings from these very warm, watered animals. And I talked about them at the Fens meeting, I had a graduate student come up to me and they were tears rolling down or
Speaker 1 00:45:25 Climate change and what's going to happen. It's going to, so the reason that worries me is because we see cryptically. So what if we just were to record control data at 11 degrees? The way we usually do, we would never know that something had changed. And so the problem that most preoccupies me now is trying to understand how all these innovations and an animal sees or goes through how they leave traces. That can be completely encrypted until you stress it. But I mean, that, that means you're getting perfectly normal looking rhythms. Um, but there are things that have changed that you don't reveal until you give another strong probation and you realize it's different. And so that I think is a big lesson for a human body, for the human brain. That if you have a child, when adult who's perfectly healthy and is not terribly badly stress, but still has lots and lots and lots of, you know, experiences, and maybe some bad experiences, you don't necessarily know that there've been some long lived changes in the brain until you come in. And this is the way I think about PTSD. You know, you come in with a very bad stressor. Someone has seemingly recover, you know, hidden they're cryptic changes, but are only revealed when you come back in with the stress or something, that's similar to the stress that evokes these, these changes.
Speaker 1 00:47:10 Yeah. So this is a problem, I think is a really, really deep problem for understanding animals in humans and how populations crashing them crash and how humans individually humans crashing don't crash. And I think, for example, just to riff, I don't think we're going to know for quite a while, what these two years with COVID crap have done Us much less that kids, but I mean, to us just they've changed all sorts of ways that we have of looking at the world.
Speaker 4 00:47:47 On the other hand though, I mean, one of the optimistic takeaways from your work is the amount, the incredible adaptability that organisms do have. Right. I was going to ask you about evolution and your view on, on the fan, you know, the fantastic mechanism or fantastic, uh, phenomenon of evolution, what you were just saying makes me think that, uh, you, you, um, are currently thinking that we're almost walking our way into, uh, a tighter constraint of the possibilities of our adaptability, right? And so maybe we started out with this really wide range of adaptability in, in a perfect environment, but then by putting these different stressors, the climate change, et cetera, it narrows the space of possibilities.
Speaker 1 00:48:32 I don't know. One of the things that we're trying to figure out how to study now is to ask the question. If you have an organism that is very resilient to stress or one probational, one, will it be necessarily more or less resilient to a different probation? So you can imagine that to be very resilient to temperature, you have to pay the price to be less resilient, to salinity or less resilient to pH. And so, um, or maybe there's a hierarchy of, you know, it's more important to be resilient to one thing than the other. So what we're trying to do now is to set up, um, we try and set up scenarios where we look at multiple stressors, multiple perturbations, and see if there's any interaction and see if there's appears to be a hard trade off or whether some animals are just good at everything and or whether or not.
Speaker 1 00:49:30 So we don't know the answer to that, but I think this goes a lot to the question that you were just asking, um, which is, are we narrowing possibilities? Are we just going down a different path? I mean, so I suspect if I had to guess from our earliest data, I would say that that the resilience to different stressors is uncorrelated. That, you know, you can't necessarily project whether an animal will be resilient to pH change on the basis of how resigned it is to temperature. Although of course, in the, in the wild, there's a correlated variables, but that's a separate issue, but, but that's one of the things that I think remains for us to try and figure out, because that is something that is relevant to how you deal with all different kinds of environmental stressors.
Speaker 4 00:50:23 Well, I know that, you know, I think it was a particularly warm winter for, and the crustaceans that you were pulling out there showed a higher resilience. They were harder to crash in some respects. Right. Um, but then you, the caveat was that you didn't know if those were just the survivors through the evolutionary, uh, funnel,
Speaker 1 00:50:43 Right? So we've had that, we've had that result sort of several times now, really warm winters giving rise to I crashed temperatures. Um, the, the worst of it is that the water temperature is sliding up. And so in 2012, um, it was an unusual result and now it's becoming more usual. So we're now starting to look at, um, we get data from Manoa Bowie, 16 miles out in the ocean for average water temperature. And we plot that against various and sundry measures.
Speaker 4 00:51:26 That's the closest you can get. Huh? Take it to knowing how their environment,
Speaker 1 00:51:31 All these guys come 60 miles out is it's not, it's probably reasonable.
Speaker 4 00:51:39 You just need to put a tracker on like slip a tracker in the fishermen's pocket. Right.
Speaker 1 00:51:43 So we actually wants to try to, um, have a colleague whose son was modeled a little computerized thing to put on his pots and he's supposed to collect for us and we'd know the temperature, we'd know when and where and everything. He brought animals in once lane disappeared with our tracker. And like I said, you know, 22 year old fishermen are not necessarily reliable.
Speaker 4 00:52:12 You can't trust them. No,
Speaker 1 00:52:17 It was not a very expensive device, but I would say he obviously got bored.
Speaker 4 00:52:22 Yeah. Oh, I see. Okay. I mean, but isn't that in some sense, the way evolution works, I mean, is this something to fret over or is it something to celebrate that there are lobsters that can survive and thrive and that this gives rise to the new breed of lobsters when our world is boiling, there'll be, they'll still be lobsters that have, you know, survived the evolutionary, uh, gauntlet.
Speaker 1 00:52:48 Um, yeah.
Speaker 4 00:52:52 It's hard for me to judge
Speaker 1 00:52:53 All of Boston's going to be underwater by then.
Speaker 4 00:52:56 Okay. Well, there'll be, there'll be closer. The crabs will be closer to you then.
Speaker 1 00:53:01 Yeah. I don't know. I it's, um, I always get nervous when I see things in the wild that are clear indications of either climate change or pollution. So for example, every couple of years we live on the waterfront. So we have an apartment which is five stories up and right out over, if we're on pilings and looks right over the Harbor. So we see what's in the Harbor a lot. So, you know, there are lots of seabirds and there's, and every now and then in may, in June, we get jellyfish and not a year or sometimes they must do. They have, we don't see very many of them every now and then there's giant infestations. And then they usually have a little beautiful way, the moon jellies, and then not this past year, but the year before with these giant really ugly red things, whose name, I can't remember, we're not supposed to see that.
Speaker 1 00:54:08 And then that same year I was looking out one day over the water and I said, holy shit, there are all these dead fish, all these dead carp in the water. And what happened is it was very warm and their carp in the Charles river and carp need a really high oxygen level in the water. And the 10, you know, there's less oxygen or water than there isn't cold water. So it was the water temperature goes up, the oxygen levels go down. And then if algae start growing the LGB, eat the oxygen. So it was this giant, giant, giant number of carp who all died at the same time and some little Washington underneath. And I look at that and I said, Hmm, this is not good. I don't like seeing dead carp in the Harbor
Speaker 4 00:55:02 Or ugly,
Speaker 1 00:55:05 Ugly, ugly, ugly. So, um, so you know, if you see it right, just like, you know, you see the giant fires in the west coast, it's not supposed to be happening like that. So you just don't like it.
Speaker 4 00:55:28 Well, of course not. Well, and you're, you're working toward helping change it. Correct. So you're doing your part. I'm not doing my part. You're doing your part.
Speaker 1 00:55:37 Yeah. What kind of car do you drive?
Speaker 4 00:55:43 It's the Colorado cockroach is what it's called.
Speaker 1 00:55:47 No, but I have a new one, which is a plug-in hybrid.
Speaker 4 00:55:51 Yeah. Great. You're taking, you're taking my parking spots up here in Colorado, but no, yeah. The plugins always get the best. Those in the handicap spots are always the best spots these days. There's a lot of, there's a lot of Teslas and electric cars out here.
Speaker 1 00:56:06 Yeah. But anyway, so that's, that's that,
Speaker 4 00:56:11 Uh, I'm just curious what you, if you have an opinion on the current hype or boom explosion, uh, in deep learning and using deep learning models to help understand, uh, brains,
Speaker 1 00:56:24 I think we'll survive it.
Speaker 4 00:56:26 Okay. Is it, it's a poisonous big red thing? In other words,
Speaker 1 00:56:32 I don't know. I mean, I, so there's this giant giant fad. Yeah. Now it will eventually win itself out. So there will be some very important advances that we will get that benefit from that way of thinking and those technologies. And they will have thrown into the universe a lot of, not very interesting papers along the way. I can't tell the difference yet. So in 10 years, you'll know what really important working from that.
Speaker 4 00:57:12 So one of the latest trends is using dynamical systems theory, you know, to talk about manifolds, um, so that you can reduce the dimension and talk about what's going on in a large population of neurons. Do you see that as a promising way forward in scaling up from something like a 30 neuron thing to a billion neuron neuron thing? Okay.
Speaker 1 00:57:33 Since I don't think well about large numbers, I, I, I can't map, I can't map those very abstract manifolds back onto hill real things. So if other people can, and it helps them. Right. But I don't know.
Speaker 4 00:57:51 Last question, Dave, I appreciate you hanging on with me here. Has your career turned out the way that you envisioned it and if not, uh, how has it been different than what you envisioned?
Speaker 1 00:58:03 I didn't actually envision it.
Speaker 4 00:58:06 Maybe that's the best way
Speaker 1 00:58:07 I just kept on going a year or two at a time. I actually had no.
Speaker 4 00:58:15 So five years from now, you don't know you're going to be doing
Speaker 1 00:58:18 Well five years from now, the three possibilities possibly number one is I'll be dead. Possibility. Two is I'll still be working or possibly three I'll have stopped working
Speaker 4 00:58:31 Climate change, um, stalwart climate change, um, leader. Okay.
Speaker 1 00:58:39 I don't have that much political activism in that, that level anymore.
Speaker 4 00:58:47 In five years, hopefully I'll be driving an electric car. So you won't get on my case about what I drive
Speaker 1 00:58:52 Well, but my husband, of course, who worries about what we have to do the environment to get the things that you need to build batteries. So in five years we may decide that the battery technology is really bad in different ways. So I guess we'd have to go find another planet
Speaker 4 00:59:13 Forward. We go. I love that solution. However, let's expand into space, but that's a, that's another hour, I suppose,
Speaker 1 00:59:21 Sad solution because it means we've just destroyed our planet.
Speaker 4 00:59:26 Well, it doesn't have to, I think we should go anyway, destroy or not destroy.
Speaker 1 00:59:30 You should go anyway, but, um, hopefully the young ones will do that.
Speaker 4 00:59:36 Thank you so much for your time. I really appreciate it.
Speaker 1 00:59:38 Thank you. Bye. Bye.
Speaker 4 00:59:45 Brain inspired is a production of me and you. I don't do advertisements. You can support the show through Patrion for a trifling amount and get access to the full versions of all the episodes. Plus bonus episodes that focus more on the cultural side, but still have science go to brain inspired.co and find the red Patrion button there to get in touch with me.
[email protected]. The music you hear is by the new year. Find
[email protected]. Thank you for your support. See you next time.
Speaker 6 01:00:21 Um, Uh,
